The energy it uses right now for full nodes and all the other infrastructure minus the energy it uses for the proof of work. If the proof of work is replaced by another proof mechanism, like proof of stake, it could be applauded.
Why is that the right amount? Does proof of work have no value at all?
If PoW provides the same security level as PoS then the marginal value of PoW is indeed zero (or negative due to the energy waste).
Very true. Is it the case that PoW provides the same security value as PoS?
Wouldn't the correct answer always be "no more than necessary / technically feasible"?
That standard is not applied to many things. There are many cases where it makes sense to use a technology that uses more energy than technically feasible because it offers other benefits. For example, using java instead of assembly.
This is actually a core principle in Ethereums community. It's called minimum viable issuance. Ethereum 2.0 is specifically designed with this principle in mind. No one wants to overpay or underpay for security, so market forces decide what level of issuance and risk is acceptable.